Mixture-of-experts routing

Vision Transformers

Superseded baseline#243 of 1,370 most-superseded

Cited as a baseline — critiqued by newer work, not yet beaten on a benchmark here

2 papers critique it · 0 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites Vision Transformers as a baseline.

these gains come at the expense of substantial computational overhead (e.g., Restormer demands 367 GFLOPs for 256×256 images), posing a bottleneck for high-resolution image processing
M2Restore: Mixture-of-Experts-based Mamba-CNN Fusion Framework for All-in-One Image Restoration
Using ViTs into LLIE pipelines has demonstrated superior performance in challenging environments; however, the high computational cost of ViTs and their tendency to overfit require architectural innovations, such as the use of lightweight modules or task-specific adaptations.
ISALux: Illumination and Segmentation Aware Transformer Employing Mixture of Experts for Low Light Image Enhancement

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.